After seeing a real example (ledge of apple orchards) about composing panoramic images, we get into our Laboratory Virtual to discuss some of the possibilities of the software “Microsoft Image Composite Editor”
To perform the various tests that subject this creation software panoramic compositions, has built a simplified model of a small part of a city. The scene was created using animation software “Blender“.
This model is meant to simulate a scenario on which we can walk virtually and so “take some pictures” (as if we were in town visiting) for retrofit.
Stage
The scenario consists of a set of buildings designed with a regular pattern of windows at different scales, that allow us to detect, if produced, “frequency errors” of images.
The first geometrical model incorporates few elements to simplify the analysis from the point of view projection systems incorporating. It is an aspect that should be affecting the computational complexity required to compose the panorama. Should be considered in evaluating results .
? By reducing the number of patterns, increases the number of combinations in comparison algorithms?
The software must be designed to overlap regions or areas of pixels of the image so, to reduce the number of colors, there are many ways overlap and thus the problem is more “variable”.
The chances of “puzzle” to be solved to compose the fewer differentiators panoramic images we use have.
In the first tests with identical buildings with colors and shapes (simpler) , images appeared in most cases did not correspond to the model. It was chaotic. Even using only basic colors and some directional lighting (Flat sin only shadows) surprising results.
City map
El assembly is formed by Blocks at different heights, arranged in a more or less regular in rows (streets), around a central zone buildings discharged.
The heights are designed to simulate different camera movements and find enough to negate variability.
Advantages and disadvantages
This working model, basado in out laboratory virtual, brings clear advantages and disadvantages.
Among the advantages we can mention:
- Comfort to take pictures from anywhere, regardless of the city traffic.
- Possibility of automated imaging and modify camera parameters such as orientation angles accurately, etc..
- Ability to perform “small” variations in the model to test specific aspects, both the geometry of the ways the colors and textures of objects.
- Comfort to set lighting or weather conditions on demand.
But the disadvantages are also present:
- It is a simulation and results may differ lesser or greater extent
- Perform a detailed model can be time-.
- It is difficult to incorporate some “human factors” that only experience allows us to appreciate
In any case it is an interesting experience that can bring results, at least, curious.
Rotation Exercise
The first exercise is to place the camera at a fixed point and turn it around for imaging. This mode can be selected in the tab “Stitch” with optional “Rotating Motion”.
The system will assume that all images were obtained from the same point, and that is a cylindrical projection type (projections will discuss the new entries).
To verify the results by contrast has made an animation in Blender, that simulates what would be captured from the same point with a video camera.
The result obtained with composition software can be compared to the previous animated image.
If you click on the image to see size.
You can change the perspective, and adjust various parameters that will be in new entries. As I leave the following example:
The first results largely validate the procedure based on the “Laboratory Virtual”, demonstrating that the tool to have a reference for the comparison of results. We will continue to deepen with more elaborate examples.
Does it seem useful working environment exposed?
Must be connected to post a comment.